1. Methodology for multispectral camera calibration using frequency component separation.
- Author
-
Juárez-Trujillo, Iván A., Zavala-de Paz, Jonny P., Isaza, César, Anaya-Rivera, Ely K., and Rizzo-Sierra, Jose Amilcar
- Abstract
The calibration process of a multispectral camera is fundamental for analyzing images where color standards must be referenced in a post-processing stage. This work presents a new method to process images acquired with a multiband/multispectral camera and scanned for reflectance information with a commercial spectrophotometer. The strategy implemented is based on a general illumination model describing lighting as an image component that changes depending on the captured environment and reflectance as a component depending only on the object's surface. Since the relationship between intensity, luminance, and reflectance in a captured image is not linear. A novel general result is that illumination adjustment generates a valid reflectance map with better surface information. An equalization of reflectance components using RGB model separation to standardize the same pixel value for each of the images was performed on eight certified Lucideon Std ceramic tiles. Lighting of 6000k was used to perceive color similar to normal daylight to each tile, both data captured in the same position. Possible industrial applications of the proposed methodology include reducing time and complexity in color inspection, multispectral image processing, and obtention of better color surface representations under various lighting conditions and application requirements. Particularly, this work shows image results interpreted in CIE RGB color space where B12 has values of 44, 61, and 63; B11 has values of 141, 176, and 84; B10 has values of 214, 176, and 203; B9 has values of 123, 89, and 82; B8 has values of 201, 66, and 225; B7 has values of 86, 73, and 103; B6 has values of 22, 37, and 160; B5 has values of 108, 107, and 64. Those compared to values measured by an spectrophotometer as reference device were in good correspondence, achieving percentage calibrations of ∼ 100 % . [ABSTRACT FROM AUTHOR]
- Published
- 2024
- Full Text
- View/download PDF